Chess_Myth_Buster

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

Hugo
Posts: 782
Joined: Tue Dec 01, 2009 11:10 am

Chess_Myth_Buster

Post by Hugo »

Hi all

after the weak performance of the new Deep Junior 13 engine in my 5 +3 gauntlet, I heared roumors about how strong DJ13 is at long time controlls.
Reason enough for me to start a match to find out a little about those myths.
Conditions :
engine match between Houdini 1.5a and Deep Junior 13.
Both engines 4 CPU, ponder ON, 4 GB hashtables.
Houdini 1.5a uses full gaviota(6GB)
Deep Junior 13 uses Nalimov 3-4-5 some 6( 40GB)
10 moves from Hiarcs book.
timecontroll 40/120min + 40/120min + 120min

games are broadcasted (not verry comfortable). Please refresh every 5 min.
http://www.clemens-keck.de/livegames/

enjoy :-)

Clemens Keck
Hugo
Posts: 782
Joined: Tue Dec 01, 2009 11:10 am

Re: Chess_Myth_Buster

Post by Hugo »

hello

game one is over, and at the end it was verry thrilling.
Houdini showed an advantage of -0.71 at move 106.
Could Junior have won this?

[Event "Chess_Myth_Buster"]
[Site "Microsoft"]
[Date "2012.02.03"]
[Round "1.1"]
[White "Houdini 1.5a x64-gtb-T4"]
[Black "Deep Junior 13 T4"]
[Result "1/2-1/2"]
[ECO "B67"]
[Annotator "0.27;0.27"]
[PlyCount "274"]
[EventDate "2012.02.03"]
[EventType "tourn"]
[TimeControl "40/7200:40/7200:7200"]

{Intel(R) Xeon(R) CPU W5580 @ 3.20GHz 3192 MHz W=28.5 plies; 9,
136kN/s; 14,638,645 TBAs; Hiarcs13hBook.ctg B=26.9 plies; 6,900kN/s; 3,200,
178 TBAs; Hiarcs13hBook.ctg} 1. e4 {B 0} c5 {B 0} 2. Nf3 {B 0} d6 {B 0} 3. d4 {
B 0} cxd4 {B 0} 4. Nxd4 {B 0} Nf6 {B 0} 5. Nc3 {B 0} a6 {B 0} 6. Bg5 {B 0} Nc6
{B 0} 7. Qd2 {B 0} e6 {B 0} 8. O-O-O {B 0} Bd7 {B 0} 9. f4 {B 0} h6 {B 0} 10.
Bh4 {B 0} b5 {B 0} 11. Bxf6 {0.27/24 554} gxf6 {0.27/21 522} 12. Kb1 {0.26/25
543 (f5)} b4 {0.27/22 194 (Qb6)} 13. Nce2 {0.41/26 285} Qb6 {0.23/22 209} 14.
Nxc6 {0.49/26 50 (g4)} Bxc6 {0.17/22 170} 15. f5 {0.37/27 271} e5 {0.32/24 98}
16. Ng3 {0.33/27 174} Qc5 {0.21/25 610} 17. Nh5 {0.30/28 0 (Bd3)} Be7 {0.12/24
345} 18. Bd3 {0.23/26 543} a5 {0.18/24 0 (d5)} 19. Qe2 {0.31/25 169} a4 {0.26/
24 623 (0-0-0)} 20. Bc4 {0.52/23 170} a3 {0.22/24 73 (Rg8)} 21. b3 {0.76/25 185
} O-O-O {0.35/24 51} 22. Qf3 {0.98/26 260} Rhf8 {0.37/24 25} 23. Ng3 {0.99/27
153} Rd7 {0.43/24 205 (Bb7)} 24. Rd3 {0.99/28 176 (Rhe1)} Kb8 {0.56/23 304} 25.
Rhd1 {0.99/28 0} Qa5 {0.62/24 252 (Qb6)} 26. h4 {1.07/26 797 (Nf1)} Qb6 {0.58/
25 185 (Bb7)} 27. Nf1 {1.04/29 213} d5 {0.58/27 249} 28. exd5 {1.03/30 395} Bb7
{0.58/27 0} 29. Nd2 {1.03/26 193 (Ng3)} Rg8 {0.58/27 569 (Rh8)} 30. Ne4 {0.80/
27 525} Bc5 {0.68/26 0} 31. Nxc5 {0.80/26 238} Qxc5 {0.66/26 0} 32. g4 {0.79/
26 173} Rc8 {0.66/26 171 (Ka7)} 33. Qg2 {0.79/27 107} Rd6 {0.66/24 0} 34. h5 {
0.79/27 112} Rg8 {0.66/26 178} 35. Qe2 {0.79/28 0} Bc8 {0.66/27 578 (Ka7)} 36.
Rf1 {0.79/29 324 (c3)} Bb7 {0.65/26 379} 37. Qf3 {0.79/30 0 (Rc1)} Kc7 {0.50/
24 243 (Ka7)} 38. Re1 {0.66/28 154 (c3)} Kd8 {0.34/25 186 (Rgd8)} 39. Qg3 {0.
57/28 205 (Kc1)} Bc8 {0.22/27 312 (Kd7)} 40. Qh3 {0.57/29 209} Bb7 {0.24/28 281
} 41. Qf3 {0.57/29 0 (Qg2)} Ke7 {0.24/25 176 (Ke8)} 42. Qf4 {0.57/27 168 (Kc1)}
Rg5 {0.21/25 223} 43. Qe4 {0.51/27 9 (Qg3)} Kf8 {0.17/26 155 (Qf2)} 44. Qe2 {
0.55/28 168 (Red1)} Rg8 {0.08/26 164 (Kg8)} 45. Red1 {0.47/28 352 (Rc1)} Kg7 {
0.08/25 112} 46. Qe3 {0.46/27 52 (Qd2)} Qa5 {0.08/27 304} 47. Kc1 {0.46/28 0
(Qf3)} Kh7 {0.08/26 237 (Rd7)} 48. Qe4 {0.40/27 267} Qc5 {0.08/28 0 (Rgd8)} 49.
Qe2 {0.40/29 193 (Kb1)} Rc8 {0.08/28 139 (Rgd8)} 50. Qg2 {0.40/29 197 (Qe3)}
Qa5 {0.08/28 161 (Rcd8)} 51. Qe4 {0.40/28 168 (R1d2)} Qa7 {0.08/28 220 (Qc5)}
52. Qe3 {0.40/29 188 (Kb1)} Qa5 {0.08/27 118} 53. Qd2 {0.40/29 76} Qc5 {0.08/
28 1077 (Rcd8)} 54. Re1 {0.35/28 186} Rd7 {0.07/26 0 (Kg7)} 55. Qg2 {0.35/28
193 (Qe3)} Qa5 {0.07/26 117} 56. Rd2 {0.35/27 86 (Qe4)} Rc5 {0.07/27 219 (Kg7)}
57. Red1 {-0.02/25 257} Ba6 {0.07/28 0 (Rxc4)} 58. Bxa6 {0.12/27 193} Qxa6 {0.
07/28 0} 59. Kb1 {-0.03/28 947 (Rd3)} Qd6 {0.00/27 155 (Rc8)} 60. Rd3 {-0.07/
27 149} Rc8 {0.00/28 63 (Kg7)} 61. Re1 {-0.07/28 157} Qc5 {0.00/27 0 (Kg7)} 62.
Rc1 {-0.07/28 128 (Kc1)} Rd6 {0.00/28 166 (Kg7)} 63. Qg1 {-0.06/28 132 (Qd2)}
Qa5 {0.00/28 136 (Qb5)} 64. Qh1 {-0.07/29 170 (Qg2)} Kg8 {0.00/28 126 (Qc5)}
65. Rcd1 {0.04/28 131 (Qe4)} Qc5 {0.00/28 123} 66. R1d2 {-0.01/29 57 (Rc1)} Kg7
{0.00/28 186} 67. Qg2 {0.00/29 39 (Qe1)} Re8 {0.00/28 191 (Rd7)} 68. Qe4 {-0.
07/28 184} Qg1+ {0.00/28 0} 69. Rd1 {-0.07/26 0} Qc5 {0.00/29 145} 70. Qe3 {-0.
07/30 220 (R1d2)} Qb5 {0.00/28 157 (Qa5)} 71. Qe1 {-0.07/29 248 (Kc1)} Kh7 {0.
00/26 254 (Red8)} 72. Qe4 {0.10/29 257} Qc5 {0.00/29 107 (Rc8)} 73. Qe3 {0.10/
30 220} Qa5 {0.00/29 0} 74. Kc1 {0.10/29 208 (R3d2)} Rg8 {0.00/29 333 (Red8)}
75. Qf3 {0.09/30 228 (Qe4)} Rgd8 {0.00/30 192} 76. Qe4 {0.07/30 94} Qc5 {0.00/
30 89} 77. Qe3 {0.09/32 473 (R1d2)} Qb5 {0.00/28 268 (Qa5)} 78. Qd2 {0.09/29
107 (Qf3)} Qc5 {0.00/30 239 (Kg7)} 79. Qg2 {0.09/29 124 (Qe3)} Re8 {0.00/30
372 (Kg8)} 80. Qg1 {0.09/30 196 (Qd2)} Qc7 {0.00/28 398 (Qa5)} 81. Qe3 {0.09/
29 144} Red8 {0.00/27 36 (Rc8)} 82. Kb1 {0.09/29 144} Qa5 {0.00/28 22 (Kg7)}
83. Qe4 {0.05/29 251} Qc5 {0.00/30 16} 84. Qg2 {0.00/29 215 (Qe3)} Kg8 {0.00/
29 261 (Re8)} 85. Kc1 {0.08/29 176} Rc8 {0.00/27 0 (Kg7)} 86. R1d2 {0.08/30
127 (Qd2)} Kg7 {0.00/28 232} 87. Qf2 {0.08/31 0 (Kd1)} Qb5 {0.00/28 197} 88.
Qe3 {0.08/30 0} Rc5 {0.00/29 163 (Rcd8)} 89. Qe4 {0.08/28 123 (Qe2)} Rd7 {0.00/
29 226} 90. Qf3 {0.08/29 0 (Kd1)} Rd6 {0.00/28 206} 91. Qg2 {0.08/29 0 (Kd1)}
Qe8 {0.00/29 267 (Rc8)} 92. Rd1 {0.08/28 126 (Qe4)} e4 {0.00/28 336 (Qb5)} 93.
Rd4 {-0.07/27 156} e3 {0.00/28 56} 94. Qe4 {-0.07/28 65} Qb5 {0.00/27 79 (Qd7)}
95. g5 {0.15/26 152 (Qxe3)} Qe2 {0.00/23 39} 96. gxf6+ {-0.19/26 193} Kxf6 {0.
00/24 0 (Kh7)} 97. Rxb4 {0.11/26 169 (Kb1)} Rcxd5 {0.02/24 55} 98. Qh4+ {0.11/
26 0} Kg7 {0.05/28 158} 99. Qg4+ {0.11/28 0} Qxg4 {0.11/29 176} 100. Rxg4+ {0.
02/31 0} Kf6 {0.11/29 131} 101. Rgg1 {-0.03/30 3 (Rxd5)} Rd8 {0.00/27 174 (Rd7)
} 102. Rde1 {-0.04/30 198} Re8 {0.00/27 0} 103. Rg3 {-0.04/30 694} e2 {0.00/29
0 (Rde5)} 104. c4 {-0.04/28 91} Rde5 {0.00/28 25} 105. Kd2 {-0.05/29 166 (Rf3)}
Kxf5 {0.00/29 180} 106. b4 {-0.03/29 0 (Rg2)} Kf4 {-0.71/26 164 (Re4)} 107.
Rxa3 {0.11/28 142 (Rd3)} Rd8+ {-1.04/25 53} 108. Rd3 {0.43/25 35} Rxd3+ {-0.46/
27 121} 109. Kxd3 {0.42/30 0} Re3+ {0.00/27 117} 110. Kd2 {0.38/30 104} Kf3 {
0.00/29 0} 111. b5 {0.38/29 1 (c5)} Re8 {0.00/28 630} 112. b6 {0.34/31 0 (c5)}
Rd8+ {0.00/28 49} 113. Kc3 {0.34/31 282} Rd1 {0.01/30 0} 114. Rxe2 {0.34/29 1}
Kxe2 {0.01/29 96} 115. c5 {0.34/30 84} Rb1 {0.01/27 13 (Rc1+)} 116. a4 {0.97/
24 23} f5 {0.42/29 66} 117. Kc2 {0.97/27 15} Rb4 {0.43/28 37} 118. a5 {0.97/28
0} f4 {0.43/27 21} 119. c6 {0.97/26 0} f3 {0.43/25 32} 120. c7 {0.97/27 0} Rc4+
{0.43/24 33} 121. Kb3 {0.97/28 0} f2 {0.43/25 72} 122. b7 {0.97/27 0} f1=Q {0.
40/25 171} 123. c8=Q {1.07/30 0} Rxc8 {0.40/24 18} 124. bxc8=Q {1.07/30 0} Qb1+
{0.00/24 33} 125. Ka3 {1.09/33 53 (Ka4)} Qd3+ {0.00/28 85 (Qa1+)} 126. Ka4 {1.
16/33 72 (Kb4)} Qe4+ {0.00/28 98 (Qd1+)} 127. Kb5 {1.16/33 6} Qb1+ {0.00/23 23}
128. Ka6 {1.25/36 326} Qd3+ {0.00/29 0} 129. Kb6 {1.25/36 78 (Ka7)} Qb3+ {0.00/
23 74} 130. Ka7 {1.25/37 0 (Ka6)} Qe3+ {0.00/29 78} 131. Kb8 {1.25/36 29 (Ka6)}
Qe5+ {0.00/26 52} 132. Qc7 {1.25/36 0} Qxh5 {0.00/26 44} 133. Qc4+ {0.00/0 0}
Ke3 {0.00/0 0} 134. Qb3+ {0.00/0 0} Ke4 {0.00/0 0} 135. Qe6+ {0.00/0 0} Kf3 {
0.00/0 0} 136. a6 {0.00/0 0} Qb5+ {0.00/0 0} 137. Kc7 {0.00/0 0} Qa5+ {0.00/0 0
} 1/2-1/2

C.K.
Martin Thoresen
Posts: 1833
Joined: Thu Jun 22, 2006 12:07 am

Re: Chess_Myth_Buster

Post by Martin Thoresen »

So you have 1 W5580?

Then with ponder on you are running your system hyperthreaded?

Just curious.
lucasart
Posts: 3241
Joined: Mon May 31, 2010 1:29 pm
Full name: lucasart

Re: Chess_Myth_Buster

Post by lucasart »

Martin Thoresen wrote:So you have 1 W5580?

Then with ponder on you are running your system hyperthreaded?

Just curious.
good point!
i assume you're using a quad core CPU?

so A and B play with 4 threads each. the problem with using pondering is that when A plays, and its 4 threads are busy, B also has 4 busy threads pondering. So you have 8 threads running on 4 CPU, which is somewhat suboptimal. It would make more sense to use 2 threads per engine if pondering, or switch pondering to off.

i generally prefer pondering off, as it makes the most sense generally. if B has no pondering feature, here what happens:
1/ A plays: 4 threads thinking, B's 4 threads are idle
2/ B plays: 4 threads thinking, A's 4 threads are pondering

1/ A uses 100% of the CPU resources
2/ both A and B use 50% resources

So pondering on introduces an unjust biais, and favors A in this situation... Which is why I think that pondering on is a silly idea. It comes from the misconception that computers, like humans, should be allowed to think on their opponent's time. But the reasons why this conception is flawed are:
1/ time doesn't affect computer's playing skills in any compararable way to humans. If computer plays in 1 second instead of 10, the quality of the moves will (on average) not be very different. time pressure is a huge factor in human games, but nowhere near as much so in computer games. perhaps computer will reach depth 12 instead of 13 if 1 second i/o 10 second, or something like that, not a big difference
2/ both programs are concurrently running on the same CPU, so when one thinks it sucks up the CPU resources of the other. 2 humans playing have 2 distinct brains, and there's no such interaction.

So I always use pondering off in my rating list for the above mentionned reasons, and I would encourage you to do the same.
lucasart
Posts: 3241
Joined: Mon May 31, 2010 1:29 pm
Full name: lucasart

Re: Chess_Myth_Buster

Post by lucasart »

the *only* case where pondering makes sense, is of course when you play against the computer. when you think the computer is pondering, and the fact that you're thinking doesn't jeopardize the computer's resources in any way.

and if you think for a long time on a move, the computer will have had more time to prepare its response, in the form of filling up its search hash table.

it also makes sense when the computer plays on a chess server, as the situation is exactly the same (the opponent whether human or computer is playing on a different machine)
beram
Posts: 1187
Joined: Wed Jan 06, 2010 3:11 pm

Re: Chess_Myth_Buster

Post by beram »

the Myth continues, third game is for DeepJunior13 now ! outstanding well played. Strange optimistic evals of 0,50 in middlegame from Houdini2.0C with his wrong placed bishop on a4. In the end the kingside of Houdini was nicely dismantled by DeepJunior.
User avatar
Dr.Wael Deeb
Posts: 9773
Joined: Wed Mar 08, 2006 8:44 pm
Location: Amman,Jordan

Re: Chess_Myth_Buster

Post by Dr.Wael Deeb »

beram wrote:the Myth continues, third game is for DeepJunior13 now ! outstanding well played. Strange optimistic evals of 0,50 in middlegame from Houdini2.0C with his wrong placed bishop on a4. In the end the kingside of Houdini was nicely dismantled by DeepJunior.
Not that I waited for this third game result but in my rating list I have the Juniors play much stronger at longer time controls....that said,it doesn't mean that I play blitz time controls but rather comparing my Junior's Elo performance to the other lists and discovering that Junior indeed needs more time to show it's real chess abilities....

Speaking of long time controls,that's 20 minutes + 20 seconds increment on an overclocked i7 machine with 1G of ram for hte hash table....
Dr.D

P.S.I haven't purchased Deep Junior 13 yet but I will certailny will in the upcoming days....
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
MM
Posts: 766
Joined: Sun Oct 16, 2011 11:25 am

Re: Chess_Myth_Buster

Post by MM »

Dr.Wael Deeb wrote:
beram wrote:the Myth continues, third game is for DeepJunior13 now ! outstanding well played. Strange optimistic evals of 0,50 in middlegame from Houdini2.0C with his wrong placed bishop on a4. In the end the kingside of Houdini was nicely dismantled by DeepJunior.
Not that I waited for this third game result but in my rating list I have the Juniors play much stronger at longer time controls....that said,it doesn't mean that I play blitz time controls but rather comparing my Junior's Elo performance to the other lists and discovering that Junior indeed needs more time to show it's real chess abilities....

Speaking of long time controls,that's 20 minutes + 20 seconds increment on an overclocked i7 machine with 1G of ram for hte hash table....
Dr.D

P.S.I haven't purchased Deep Junior 13 yet but I will certailny will in the upcoming days....
Hi Doc,

i suggest you to wait till the next CCRL and CEGT. No need to hurry...ccrl has 40/40 and/ or at least cegt has 40/20.

Kind Regards
MM
User avatar
Dr.Wael Deeb
Posts: 9773
Joined: Wed Mar 08, 2006 8:44 pm
Location: Amman,Jordan

Re: Chess_Myth_Buster

Post by Dr.Wael Deeb »

MM wrote:
Dr.Wael Deeb wrote:
beram wrote:the Myth continues, third game is for DeepJunior13 now ! outstanding well played. Strange optimistic evals of 0,50 in middlegame from Houdini2.0C with his wrong placed bishop on a4. In the end the kingside of Houdini was nicely dismantled by DeepJunior.
Not that I waited for this third game result but in my rating list I have the Juniors play much stronger at longer time controls....that said,it doesn't mean that I play blitz time controls but rather comparing my Junior's Elo performance to the other lists and discovering that Junior indeed needs more time to show it's real chess abilities....

Speaking of long time controls,that's 20 minutes + 20 seconds increment on an overclocked i7 machine with 1G of ram for hte hash table....
Dr.D

P.S.I haven't purchased Deep Junior 13 yet but I will certailny will in the upcoming days....
Hi Doc,

i suggest you to wait till the next CCRL and CEGT. No need to hurry...ccrl has 40/40 and/ or at least cegt has 40/20.

Kind Regards
Thanks buddy...I'll listen to your advice :D
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
tomgdrums
Posts: 736
Joined: Wed Dec 02, 2009 9:48 am

Re: Chess_Myth_Buster

Post by tomgdrums »

I have only been able to check once in awhile but the positions in these games have looked really interesting!

A good match so far (so it seems)